Understanding the interaction determinants of CAPN1 inhibition by CAST4 from bovines using molecular modeling techniques

Abstract

HCV-induced CAPN activation and its effects on virus-infected cells in a host-immune system have been studied recently. It has been shown that the HCV-nonstructural 5A protein acts as both an inducer and a substrate for host CAPN protease; it participates in suppressing the TNF-α-induced apoptosis response and downstream IFN-induced antiviral processes. However, little is known regarding the disturbance of antiviral responses generated by bovine CAPN activation by BVDV, which is a surrogate model of HCV and is one of the most destructive diseases leading to great economic losses in cattle herds worldwide. This is also thought to be associated with the effects of either small CAPN inhibitors or the natural inhibitor CAST. They mainly bind to the binding site of CAPN substrate proteins and competitively inhibit the binding of the enzyme substrates to possibly defend against the two viruses (HCV and BVDV) for anti-viral immunity. To devise a new stratagem to discover lead candidates for an anti-BVDV drug, we first attempted to understand the bovine CAPN-CAST interaction sites and the interaction constraints of local binding architectures, were well reflected in the geometry between the pharmacophore features and its shape constraints identified using our modeled bovine CAPN1/CAST4 complex structures. We propose a computer-aided molecular design of an anti-BVDV drug as a mimetic CAST inhibitor to develop a rule-based screening function for adjusting the puzzle of relationship between bovine CAPN1 and the BVDV nonstructural proteins from all of the data obtained in the study.

Figure 1

The hydrophobic interaction surfaces of the binding-site residues near the location at which the local kink conformer (residues 611–624 of rat CAST4 and 678–691 of bovine CAST4) binds. (a) Close-up views of the residues (Leu612 and Ile619) of CAST4 binding at the key contacts of rat CAPN2; and (b) the corresponding residues (Leu679 and Ile686) bound to the active site of bovine CAPN1. Both the subdomains B of CAST4 assume a similar backbone conformation for the distorted local kink. These local kink conformers show the most similar patterns of hydrophobic contact distribution across the CAPN subgroups. Hydrogen bonds are represented by green dashed lines.

Figure 2

The inhibitor specificity of the residue preferences observed for each position in subdomain B of bovine CAST4. The Leu679 mutant shows preferences for the residues Arg, Phe, Leu, and Tyr (had stabilizing effects), whereas the residues Tyr, Trp, Phe, and Arg of the Tyr690 mutant had natural effects of the almost apparent wild-type complex (from most to least stabilizing). Those mutants show distinct preferences for residues bearing hydrophobic side chains (rather than hydrophobic ones) indicating specificity for both positions. In the single-point mutation study, the mutant conformer contributes considerably to the difference in mutation binding energy from the wild-type complexes. The energy difference of each mutation on binding affinity is the difference between the binding free energy in mutated and wild-type proteins, which are predicted to cause destabilizing effects such that mutation energies greater than 0.5 kcal/mol are designated as destabilizing.

Figure 3

The π-π stacking interactions between two proline residues (Pro687 and Pro688) of CAST4 and a key tryptophan residue, Trp298 of CAPN1 from bovine in which Trp298 moved to tuck into a hydrophobic patch formed by the Ca²⁺-binding induced rearrangement of the gating loops (residues 96–108 in DI and 251–271 in DII).

Figure 4

Detailed view of the interaction sites of the Lys689 and Tyr690 residues of subdomain B, which are in contact with the entryway of the protease core DI and DII (shown in ribbon represented in blue and brown, respectively) and were mutated to alanine to reduce the size of its side-chain, rendering CAST4 less space-filling mode from the active site.

Figure 5

Inhibition profiles of its native inhibitor CAST4. The conserved KLGERDDTIPPKYQ motif in the bovine CAST4 (residues 678–691) was screened using a computational, alanine scanning mutagenesis. The results are shown as the residue preference from a wild-type complex with the corresponding energy effect (kcal/mol) of alanine mutation at the single-point position for the stability (green) and binding affinity (red) of the bovine CAPN1/CAST4 complex. Each bar chart of mutation energy is created to allow for comparing the different effects of mutation, when the mutation effects are defined as “stabilization” (mutation energy less than −0.5 kcal/mol), “natural” (mutation energy between −0.5 and 0.5 kcal/mol), and “destabilizing” (mutation energy greater than 0.5 kcal/mol).

Figure 6

Small molecule inhibitor interaction diagrams for the binding pocket of the mammalian CAPN subunits (PDB code: 1TL9, 1NX3). For both inhibitor-bound structures, inhibitors are shown as sticks accompanied by key residue representation of 2D-CAPN interacting interface. The complex structures reveal the defining their physicochemical properties for CAPN selectivity and specificity: (a) In the complex structure between rat CAPN1 and leupeptin (1TL9) at the protease core, extensive interactions help stabilize leupeptin at the active site by eight hydrogen bonds with the side-chains of Glu72, Glu261, Gln109, Cys115 residues (a blue dashed arrow) and backbone of Gly208, Gly271 residues (a green dashed arrow) and hydrophobic interactions with Leu260, Ser251, Ala273, Asn253, Ile264 residues. The domain-leupeptin complex overlaps with those of the CAPN/CAST complex structure (PDB code 3DF0) as main chain atoms r.m.s.d of 1.4 Å [38]; (b) The interaction site for the crystal complex structure (1NX3) of CAPN domain DVI (of pig) and its inhibitor PD150606 bond in a hydrophobic pocket (Val125, Leu132, Phe137, Ile169, Gln173, Phe224) of DVI where it makes favorable π-π interaction with the side-chain of His129.

Figure 7

Interaction residues associated with two LGMD2A-related mutants (Arg385His, Asp600Gly).

Figure 8

Sequence alignment of bovine CAPN1 and CAPN2 large catalytic subunit. The accession numbers for these sequences are NP_776684.1 (NCBI reference sequence) and AAI34527.1 (GenBank), respectively. Bovine CAPN1 has 81.5% sequence homology to the subunit of CAPN2. The trace residues that made up the class-specificity of CAPN in the interaction sites with subdomains A and B of the CAST4 group are indicated in red. All positions are described in terms of the bovine CAPN1 amino acid sequence. For comparison, two bovine CAPN isoforms had similar specificity toward the enzyme CAST4.

Figure 9

Interaction sites within the protease core by residue character: class-specific (blue; Ile83, Lys174, Val176, Ile254, Ser255, Asp259, Ala262, Val269, Glu300, Arg347) and bovine only species-specific (red; Ser209, Asp253, Ser256, Met260, Val263, Lys270) trace residues among key interaction contacts made the CAPN1/CAST4 complex. Notably, not adjacent regions of the catalytic Cys115 residue, the positions of trace residues from bovine CAPN1 are location-focused within the open conformation of the flexible loop of domain DII (residues 251–271) which gates active site, consistent with their binding site of conserved KLGERDDTIPPXYX motif of CAST4 subdomain B having similar inhibitory preferences in the active site cleft, but will likely not be identical against two CAPN isoforms (CAPN1 and CAPN2) from the bovine.

Figure 10

Spatial aggregation propensity (SAP) for bovine CAPN1: (a) Values of SAP at R = 10 Å which define SAP based atoms within radius-10 Å from a given atom for the catalytic subunit (domains DI-DIV) of active bovine CAPN1, along with peaks of chosen regions of the enzyme known to interact with inter-domains or other proteins; (b) SAP values (R = 10 Å) mapped onto the active bovine CAPN1 model. Positive SAP values are red (hydrophobic) whereas negative values are blue (hydrophilic); therefore, a highly exposed hydrophobic fragment would be deep red and a highly exposed hydrophilic fragment would be deep blue.

Figure 11

Overview of the 3D-complex model structure of bovine CAPN1 bound to CAST4. The modeled structures of the bovine CAPN1/CAST4 complex all have the same domain color schemes in the presence of Ca²⁺ ions: DI (blue, residues 30–220), DII (brown, residues 221–386), DIII (yellow, residues 387–543), DIV (green, residues 544–716) in the catalytic subunit of bovine CAPN when the CAST subdomains A–B and the three Ca²⁺ ions are shown as ribbon diagram and sky blue spheres, respectively. The domain DVI of small subunit (colored in pink) had not been characterized by our homology modeling and has been proposed as a homo-dimer model of the CAPN1 using the crystal structure of CAPN domain VI in PDB code 1DVI (Blanchard et al. [42]).

Figure 12

Contribution of only interaction interface residues to net binding energy in the CAPN1/CAST4 complex. Difference in binding free energy between alanine-substituted and wild-type of the bovine CAPN1 ( $\text{∆∆}{G}_{\left(mut\right)}^{binding})$ at contact residues is represented by chart bars. In the interfaces, the majority of residues of bovine CAPN1 are predicted to cause destabilizing effects (mutation energy greater than 0.5 kcal/mol is designated as destabilizing), while negative values indicate that the binding affinity increased when the side-chain was substituted by alanine.

Figure 13

The protease core (left) between the domains DI and DII and hydrophobic pocket (right) in the domain DIV of the new CAPN1-based pramacophore models generated from the modeled bovine CAPN1/CAST4 system, superimposed on the key residues of interaction interfaces. The following pharmacophore features on the key residues (Gln109, Gly113, Lys171, and Trp298 within the protease core and Lys578, His579, Arg627, and Trp616 in the hydrophobic core of DIV) are color coded, adding location constraints to the pharmacophore features, defining the relative positions of the features required for the CAST4 inhibitor to map: H-bond donor (pointed green ball), Hydrophobic (pointed cyan ball) excluded volume (grey ball), negative charge (pointed blue ball), shape constraints (grey shape) of the interaction interface.

Figure 14

Side view of the superimposed interactions interface for the rat CAPN1proteolytic core inactivated by both ZLAK-3001 and SNJ-a945, α-ketoamide-based inhibitors-bound X-ray crystal structures in the PDB code 2R9C (red) and 2G8J (green), automatically created two-dimensional diagrams of complex with PoseView [46].

Figure 15

Stereo views of the hydrophobic surface plots within CAPN1 with the hydrophobic inhibitors; subdomain A of CAST4 (the natural CAPN inhibitor, Left) and PD150606 (the small inhibitor, Right) binding sites. The inhibitors bound to the hydrophobic pocket formed by EF-hands 1 and 2, that exposed on water molecules when both non-Ca²⁺-binding and Ca²⁺-binding. The inhibitors are positioned in almost the same place assuming a similar magnitude of Ca²⁺-induced conformational changes before binding to CAPN1.